# Marcus Treadway, Staff Engineer at Fieldline Health (62 people) — read of learns-yo, May 20 2026

> 9 years writing TypeScript, currently the person who reviews every spec before it goes to a junior. Two daughters, 7 and 4. Builds furniture on weekends when he actually has weekends.

## How I got here

Someone in the Cursor Discord dropped it as a "decent attempt at the codebase-aware generation problem." I clicked because I'd already tried four tools this quarter that promised something close to this and none of them knew what `@/lib/db` meant. I had 8 minutes before standup.

## What I clicked first

The code sample. Immediately. That is always where I go because it is the only honest part of a page like this.

And the sample is actually good. Specific. `import { db } from "@/lib/db"` with an aliased path, `NotFoundError` as a typed throw, `logger.info("user.access", { id })`. That is not Stack Overflow output. Someone who writes TypeScript for a living wrote that example. The line "your db client, your logger, your conventions, every time" is doing real work here.

## Where I paused

The 200k-token context window claim: "Up to 200k tokens of repo context per translation." We have a monorepo. If that number is real and not marketing for a cherry-picked repo, this might actually thread the needle where Cursor and Copilot fall flat. I stopped and did the math on what 200k tokens covers. It is not the whole repo, but it might be enough for a service layer. I kept reading.

## What I distrusted

The bottom half of this page is selling me something completely different than the top half.

The hero says "For TypeScript shops shipping from spec docs and Notion drafts every week." Fine. I'm reading.

Then I hit this: "Who this is for: Tutors with $500K+ revenue, training-program operators, certification course creators with a list." That is not me. That is not anyone I know. And "anyone who wants to launch this in under 4 weeks (it's a heavier build)" -- launch WHAT?

Then I see "$5 to unlock the dossier" and "$99 to adopt the build." And "1 in 9 meaningful-success odds." And "Year-1 take-home: $-20,990."

This page is not a product for me to use. This is a product idea being sold to someone who wants to start a company. The TypeScript shop framing in the hero is describing the END customer, not me. I am reading a pitch for operators, not engineers.

That is a real confusion. I wasted 4 minutes before I understood what I was actually looking at.

## What would convince me

If this is a real tool I can use: a 90-second video of someone pasting a markdown spec for an actual endpoint (not a hello-world getUserById) and getting back a diff I would merge. Real repo, real conventions, real output. Not polished.

If this is a business opportunity for me to build: I would want to see one design partner quote. One sentence from one actual TypeScript team lead saying the output was closer to their conventions than what GitHub Copilot produced. Not a score. Not a Fermi estimate. One human.

The honest scoring (68/100, the negative take-home number) is actually the most interesting thing on the page. That is a weird thing to lead with and I respect it. But it also tells me no one has used this yet.

## What I'd ask in an email reply

1. The code sample uses `db.users.findUnique` which is Prisma syntax. Does the indexer actually detect the ORM in the repo and adapt, or do I need to tell it? Because we're on Drizzle and most demos like this fall apart the second you're not on Prisma.

2. When you say "indexes your codebase," does that mean a one-time snapshot I re-run manually, or does it stay in sync with git? Because our conventions drift and the whole promise breaks if the index goes stale.

3. The page says "pre-revenue, taking design partners." What does that actually mean for delivery? Are you handing me a codebase starter and a Notion doc, or is there a working API I can hit today?

## Verdict: on-the-fence

The top of this page is solving a real problem and the code sample shows genuine craft. But I genuinely could not tell, for the first five minutes of reading, whether this was a SaaS tool or an idea kit. That confusion alone would cause most engineers to close the tab. I didn't, but only because I was specifically looking for this problem.

---
*Memo by skeptic persona, generated 2026-05-20. Studio breaks own self-grading loop.*